Image

Confirmatory Factor Analysis (CFA)

Abstract

Confirmatory factor analysis (CFA) is a statistical technique used to validate the factor structure of a set of observed variables. It tests whether measured indicators reliably represent underlying latent constructs (Hair et al., 2018).
As demonstrated in the tutorial article on CFA/CB-SEM in SmartPLS by Hair et al. (2025), SmartPLS offers CFA capabilities as part of its covariance-based structural equation modeling (CB-SEM) functionality. With graphical model building and estimation based on maximum likelihood (ML), SmartPLS enables researchers to rigorously assess hypotheses about the relationships between observed and latent variables.
SmartPLS thus serves as a clear alternative to IBM SPSS Amos.
The following screenshot shows CFA results in SmartPLS for Kline’s (2023) textbook example on job satisfaction:
Kline's CFA Textbook Example in the Context of Job Satisfaction in SmartPLS

CB-SEM Algorithm Settings for CFA in SmartPLS

Maximum iterations

Defines the maximum number of iterations the optimizer will perform. This should be high enough to ensure convergence.
  • Default value: 1,000

Starting value strategy

  • Apply configured starting values
    Uses user-specified starting values if enabled. Otherwise, default starting values are applied.
  • Default strategy
    Mimics Lavaan’s defaults:
    • Loadings: Fabin-style estimates
    • Path coefficients & covariances: 0.0
    • Residual variances: 0.5 × indicator variance
    • Latent variances: 0.05
  • One-zero strategy
    A simpler approach with:
    • Loadings: 1.0
    • Variances: 1.0
    • Path coefficients & covariances: 0.0

Stop criteria

  • Gradient criterion
    Optimizer stops when:
    ||g|| < stop criterion × max(1, ||x||)
    • Default value: 10^-6
  • Function value criterion
    Optimizer stops when the improvement in the ML objective function is negligible:
    (f' – f) / f < stop criterion
    • Default value: 10^-9

Special assumptions

  • Imply latent variable correlations
    Estimates correlations between all exogenous latent variables, even without a correlation arrow.
  • Imply causal indicator correlations per construct
    Estimates correlations between causal indicators of a latent variable, even without correlation arrows.
  • Imply a variance of 1.0 for causal indicators
    Constrains all causal indicator variances to 1.0 (overrides user-specified values). Helps mimic Lavaan defaults.

Mean Structure

Most CFA and SEM analyses focus on modeling the covariance structure of observed variables. In some cases (e.g., latent growth curve modeling), including a mean structure is necessary. A mean structure involves means and intercepts of latent and observed variables and requires constraints for identification (since only p observed means are available, with p = number of observed variables).

Options

  • No mean structure (default)
    Ignores means and estimates only covariances.
  • Estimate mean structure, fix factor means to zero
    Includes mean structure. Factor means are constrained to zero, while observed intercepts are estimated freely.
  • Estimate mean structure with only user-specified constraints
    Includes mean structure with no predefined constraints. The user must specify identification constraints.

CFA Examples in SmartPLS

SmartPLS provides directly computable CFA examples from established textbooks (Byrne, 2016; Hair et al., 2018; Kline, 2023; Schumacker & Lomax, 2010). SmartPLS replicates the results reported in these sources. Try out the CFA example projects in SmartPLS!

References

Cite correctly

Please always cite the use of SmartPLS!

Ringle, Christian M., Wende, Sven, & Becker, Jan-Michael. (2024). SmartPLS 4. Bönningstedt: SmartPLS. Retrieved from https://www.smartpls.com